Machine Learning services in AWS
When we have a task that we can clearly solve using Artificial Intelligence, it is always tempting to start from scratch. If you know Amazon Web Services (AWS) you can pay for a virtual server and install all necessary libraries or directly start with an AMI with all packages preinstalled. After you lose your mind over a week after realizing you have no idea where to start, you probably start thinkin: I wish this functionality already existed.
What is Machine Learning
You have probably read the term Machine Learning in the news. It is the new buzzword. Many companies are starting to…
Come on! We are talking about AWS of course what you are doing already exists. You have many Machine Learning services available to be used both by experts and noobs.
If you know what you are doing, you should probably use SageMaker. It is basically a service that allows you to use the full potential of the cloud infrastructure. You just need to use your Machine Learning skills, and AWS takes care of the rest.
You can choose the type of instance you want to train your algorithms on (GPU…) and optimized algorithms.
With the use of Docker it provides a standard way to code your project.
If you are not a programming Jedi don’t worry. AWS provides many services where do you not have to write a single line of code (at least for the basics).
If you have to do Natural Language Processing (NLP) tasks such as sentiment analysis, entity recognition, document classifiction or topic modeling; this is your service.
When you need to translate text. Works exactly the same as Google Translator. You just indicate the source and target languages, insert the text and it gets translated.
Speech to text. Converts audio to text, thats it. Of course it has its technical complexity but it is very easy to use.
Text to speech. Like Transcribe but the other way. Converts text to audio. Some of the main characteristics of Polly:
- Allows to word customization to define its pronunciation.
- Selection of different voices.
- Code when a sentence starts and ends.
The computer vision service. It provides many functionalities:
- Object Detection.
- Image moderation. Identify offensive content.
- Facial analysis. Detects faces in images and provides information such as age or sentiment.
- Famous recognition. Identify famous people in photos.
- Text detection.
Time series forecasting. Basically an AutoML which looks for the best algorithm to predict.
The service you need if you want to build a chatbot. Works similarly to Alexa, but with less features. You have to program orders.
Amazon allows us to use their ownRecommender System algorithm with this service.
A service that provides OCR functionalities. Contrary to Rekognition, that also has OCR capabilities, Textract has been designed only for this task. You can obtain text both from images and PDFs.
This is a really interesting service. It is focused on people that would like to get started in the Reinforcement Learninf field. It consists on a graphical environment where you can train an agent to compete in car races.
Another educational service. AWS provides you with a physical camera so you can build your computer vision service.
It integrates with services such as Rekognition, SageMaker or Polly.
Finally, DeepComposer also provides a physical device, a keyboard.
The main idea is what you can just start a melody and an IA will generate the rest.
If you end up using any of this services let me know!